Re: [PATCH v18 05/23] KVM: arm64: Track the type of VM in kvm_arch

[Suzuki K Poulose]

On 18/09/2026 10:14, Fuad Tabba wrote:
> Hi Suzuki,
>
> On <date>, Suzuki K Poulose <suzuki.poulose@xxxxxxx> wrote:
> [...]
> > +       if (WARN_ON_ONCE(!kvm))
> > +               return false;
> > +
> > +       /* Cover both pKVM host and Realm VMs */
> > +       if (!kvm_vm_hyp_is_distrusting(kvm))
> > +               return true;
>
> I'd drop the WARN_ON_ONCE. 

I would say, we need it. i.e., we are filtering kvm_arch_vm_ioctl(),
where it is called from the kvm instance ioctl. So, kvm must be !NULL.


> And this early return needs the `>=` from
> patch 13, or non-protected pKVM VMs skip the filter.

Of course, This snippet is for Patch 20 and I have fixed the comparison.


> > I resorted to is_protected_kvm_enabled() to make it faster than using
> > the vm_flavor checks. I could rever those anyways.
>
> Agreed it's the faster test. I'd just call it by name at the three
> sites rather than through a macro.
>
> > Do you mean something like :
> >
> >          if (kvm_vm_is_protected_pkvm(host_kvm))
> >                  hyp_vm->kvm.arch.vm_flavor = VM_PROTECTED_PKVM;
> >          else
> >                  hyp_vm->kvm.arch.vm_flavor = VM_PKVM;
>
> Yes, exactly that.
>
> > Do we additionally need to WARN_ON() if we encounter a VM_REALM ?
>
> No, the host shouldn't be able to trigger a WARN at EL2.

Alright.

> > Apologies, I thought this one didn't change, but later patches changes
> > them, which is why I kept this.
>
> No worries. With those two in, please keep the tag.

Thanks
Suzuki

> Cheers,
> /fuad